Aphasia, an acquired language impairment affecting some or all language modalities, occurs in about 30% of people who have stroke. Behavioral aphasia treatment applied in the chronic phase of recovery is generally effective in improving language performance, but outcomes are highly variable and this variability is poorly understood. Recent advances in neuroimaging methodology have revitalized interest in the role of white matter tracts for understanding brain- behavior relationships in aphasia and for understanding treatment effects. One particularly difficult problem that has complicated efforts to visualize structural connectivity is the crossing of fiber pathways. High Definition Fiber Tracking (HDFT) is a novel approach consisting of processing, reconstruction, and tractography methods that can visualize several hundred thousand fibers from cortex and subcortex through complex fiber crossings with at least millimeter resolution. This proposal plans to use HDFT in patients with chronic aphasia following stroke, in order to study the relationships between white matter pathways and specific aspects of language behavior, as well as neuroplastic changes associated with intensive behavioral treatment. The intensive speech therapy to be used is based on current psycholinguistic concept that word production activates both lexical-semantic (word meaning) and phonological (word sound) representations. The main hypothesis is that by correlating specific white matter disconnections with semantic and phonological deficits, as well as with the response to intensive behavioral therapy, the structural correlates of language function and its recovery can be defined and the recovery potential of stroke patients can be predicted. Specifically, the proposed research will determine: the associations between damage to particular fiber tracts and functional naming deficit patterns; neuroplastic changes in language related fiber tracts induced by intensive aphasia therapy; and the value of specific fiber disconnections for predicting treatment outcomes. It is expected that the results will be consistent with the view that that the dorsal stream, defined by the arcuate and superior longitudinal fascicles, contributes to both semantic and phonological processing for word production and that the ventral stream, consisting of the uncinate fasciculus and the inferior occipital orbitofrontal fasciculus, plays a secondary role and contributes only to semantic aspects of word production. It is also anticipated that the intensity of structural neuroplastic changes will be correlated with clinical improvement; and the degree and location of neuroplastic changes will correlate with degree and type of language improvement (i.e. semantic versus phonological).